When analyzing customer feedback there are three considerations which must be top of mind: the value of information declines with time, each individual is actually speaking for many people, and don’t look at your information without seriously considering the best way(s) to first segment the respondents.
Time Value of Information
If someone detects a fire in a building but waits until the next day to call 911, you would think they are idiots, antisocial or possibly arsonists. After all, we all know that the sooner the Fire Department is called the more likely the fire will be controlled with minimum damage or loss of life.
On a more grand scale, think about the Normandy invasion by the Allies on June 6, 1944. The Allies established a phony Army, commanded by the real General George Patton, near the Southeast end of England. The purpose of the Army was to convince General Rommel, the German commander of the defense forces, that the Normandy landing was a distraction and that the main landing was at the Pas-de-Calais. It took Rommel until the end of June 7 to figure out that Normandy was the real deal and by then it was too late to effectively redeploy his reserve Army. The outcome of the war was just a matter of time.
The same phenomenon happens with feedback. If a customer complains about her mobile network and the company waits weeks or even months to respond, there is a strong likelihood that she will no longer care – she switched carriers. This graph, included by permission of Tibco Spotify, clearly explains the time value of data:
Each Respondent is Speaking For Many Customers
I was recently working with a large medical distribution company and was helping analyze their survey data. About 4500 customers, out of about 90,000 that interacted with the shipping function, completed our survey. The results were statistically valid with a margin of error of ±1.42% at a 95% confidence level. As I was thinking about the story I was getting from the data I realized that each respondent actually represented 20 customers – 90,000/4500 (themselves and 19 others). So, when 20 people said that their last order was missing an item there were about 380 other people who had exactly the same experience.
While this simple observation seems to be obvious, the people I generally talk with never internalize the fact that significant numbers of their customers (i.e., real people) have problems; they talk about “invisible” statics. Never lose sight of the fact that customers are people who feel real pain and many real decisions, which will impact your business for years to come.
Always Segment Your Data Before Drawing Conclusions
When I first learned about segmenting customers it was always Tier 1, 2 and 3 based on annual purchases. Since I knew about Pareto’s Law and ABC inventory management this approach made sense to me. Then, when I moved into managing a Customer Service profit center, I started segmenting my customers based on their service needs. For example, 24 x 7 support, 8 to 5 Monday to Friday, self-service or I’ll call when I need your help. This taught me about designing services for different segments of my customer base and helped my grow both revenue and customer satisfaction.
Now, when I talk to B2C clients we talk about segmenting customers by demographics as well as value to the business and needs. For example, age, gender, education level, income level, part of country where they live, and digital comfort level just to name a few. Leading retailers actually focus on individuals and create custom offers to address individual needs, which is where the world of Big Data interacts with business.
Even if you are a small to midsize business, you can break your customer base into manageable, homogeneous groups and create products, services, and loyalty programs just for them. As you also consider overall importance to your business you should identify the top 1, 10, or 100 customers whose departure would dramatically hurt your business (your key accounts). Then create and implement programs specifically designed to earn their long-term loyalty.
In summary, when looking at customer feedback:
- Remember that each response represents many more than one person
- Consider how quickly you can and should analyze and act on your customer feedback
- Don’t just lump all your data into one bucket but group people into groups that will allow you to support many people with one action